Diploids of the Valeriana officinalis group (Valerianaceae) in Central Europe, and an attempt to unravel the nomenclatural chaos

Abstract: Diploid populations of Valeriana officinalis L. (Valerianaceae) in Central Europe exhibit an extensive variation, with two conspicuous morphotypes. One, corresponding to the lectotype of V. officinalis, is larger, with broader, distinctly dentate leaflets, the other is in many respects similar to V. pratensis Dierb. and V. stolonifera Czern., but is glabrous, with narrow, usually entire leaflets. The two forms also differ in their ecological optimum and in geographical distribution, but are linked with numerous intermediates. Morphometric analysis (PCA) supports the picture outlined above. Nomenclatural analysis shows that the correct name for the taxon with narrow leaflets at the rank of variety is V. officinalis var. tenuifolia Vahl, with a very old synonym at the rank of subspecies, V. officinalis subsp. tenuifolia (Vahl) Schübler & Martens. A lectotype and epitype (from a plant with 2n = 14) are designated for this name. The epithet tenuifolia cannot be used for what has usually been called V. collina auct., or correctly, V. stolonifera subsp. angustifolia Soó. Further names were also studied in detail: V. officinalis var. angustifolia Wahlenb., V. officinalis var. angustifolia Hayne (for which a lectotype is designated), V. altissima Besser and V. officinalis var. altissima W. D. J. Koch. Citation: Kirschner J. & Zeisek V. 2017: Diploids of the Valeriana officinalis group (Valerianaceae) in Central Europe, and an attempt to unravel the nomenclatural chaos. — Willdenowia 47: 189–201. doi: https://doi.org/10.3372/wi.47.47301 Version of record first published online on 25 October 2017 ahead of inclusion in December 2017 issue.


Introduction
The group of taxa around Valeriana officinalis L. includes a number of names used without appropriate nomenclatural evaluation, typification and interpretation. Although the group was subject to extensive field research and several taxonomic revisions, taxonomists did not reach any reasonable consensus about the number, circumscription and rank of taxa to be recognized. Objective reasons for this situation should be sought in the rather low level of structural differentiation among taxa, the existence of polyploid series (the most important works on V. officinalis polyploi-dy include Skalińska 1947;Titz 1969;Hidalgo & Vallès 2012;Bressler & al. 2017; and other relevant works cited below) and possible occasional intermediates or hybrids. Many problems, however, stem from uneven geographical exploration of the V. officinalis group and from very different methods used to understand the population variability (Walther 1949;Voroshilov 1959;Titz & Titz 1982;Titz 1984;Holub & Kirschner 1997;Buttler & al. 2008).
Taxonomic evaluation of the group of Valeriana offi cinalis therefore remains to be completed using modern methods of genomic analyses and on a geographically representative selection of population samples. The Central

190
Kirschner & Zeisek: Diploids of the Valeriana officinalis group in Central Europe Table 1. Taxonomic summary of the Valeriana officinalis group in Central Europe. Nomenclatural citations are given in Buttler & al. (2008) and in the Euro+Med PlantBase (Kirschner & Raab-Straube 2017+).  (1946) and Buttler & al. (2008), we treat the type of V. excelsa as identical with what is called V. excelsa subsp. procurrens in Holub & Kirschner (1997) and Kirschner (2002 (1977), daß es sich um frühblühende (und stockbildende!) 'exaltata' handelt." It was Sebald (1996) who summarized the problem of variation of the diploid Valeriana officinalis. Within the diploid V. officinalis, he recognized two forms ("exal tata" and "pseudopratensis"), and their characterization is so useful that we can quote from his identification key to the Baden-Württemberg forms (translated from German, Sebald 1996: 27 European forms of this group have been studied profoundly by several specialists, most importantly by Titz & Titz (1982) and Titz (1984), and nomenclatural aspects have been outlined by Buttler & al. (2008). Results of detailed field observations in Baden-Württemberg were excellently summarized by Sebald (1996). From the above works and our own older research (Holub & Kirschner 1997;Kirschner 2002), we can derive a taxonomic survey of this group in Central Europe (Table 1). The present paper is focused on the taxonomy and nomenclature of diploids of the Valeriana officinalis group in Central Europe. We do not include diploid forms occurring in S Europe because there are several diploid entities similar to V. stolonifera that remain to be studied and properly understood (Titz & Titz 1982 not only mention the "illyrica" diploid form from the Balkans but also a diploid from the Pyrenees; the latter might be close to V. hispidula Boiss.). Seemingly, the study of Central European diploids is a relatively simple task because most of the above literature sources recognize only a single diploid taxon there, which is V. officinalis L. s.str., its name having recently been typified and interpreted (Kirschner 2007, with the name V. exaltata J. C. Mikan ex Pohl as a synonym). However, some comments in the literature indicate that another taxon might be involved in the relatively extensive variation of V. officinalis at the diploid level. Titz & Titz (1981) spoke about variants of V. exaltata with narrow leaflets in the region of Bodensee: "[shortened] Willdenowia 47 -2017 It should be added that the "pseudopratensis" form usually has leaflets entire, very narrow and, as Sebald (1996) correctly says, at the sites where it is in contact with Vale riana officinalis s.str., there are numerous intermediates, both morphological and phenological.
In the present paper we therefore aim at the characterization of the above (so-called) "pseudopratensis" form, including its general distribution, and we revise the relevant nomenclature, with the emphasis on the correct interpretation of the name Valeriana officinalis var. tenui folia Vahl.

Material and methods
Nomenclatural analysis was performed in strict accordance with the latest version of the International Code of Nomenclature for algae, fungi, and plants (hereafter "Code"; McNeill & al. 2012). Voucher specimens for chromosome counts and other material are deposited at PRA. A number of relevant herbarium collections were consulted in the course of the present study: BM, BRNM, BRNU, C, K, LI, OL, PR, PRC, W and WU. Herbarium codes are according to Thiers (2017+).
Chromosome numbers were determined in roots of mature cultivated plants, according to the techniques described in Štěpánek & al. (2011). The inevitable prerequisite for any study of the group of Valeriana officinalis was the typification and interpretation of this Linnaean name, which was done by Kirschner (2007; see also Kirschner & al. 2007).
Morphometric analyses were done on a selection of our population samples and additional herbarium specimens (LE, PRA, TU). The characters recorded were: (1) plant height at full flowering; (2) leaflet width (middle leaflet of middle cauline leaf); (3) leaflet length; (4) proportion of leaflets with teeth on both margins; (5) number of teeth (middle leaflet of middle cauline leaf); (6) number of leaflets (middle cauline leaf); and (7) proportion of leaflets with teeth (middle cauline leaf). The material measured is given in Table 2; altogether 80 specimens were measured. The material was selected to make it possible to evaluate the large-scale variation (Central Europe versus Estonia and NW Russia) and the variation within a series of large populations in C Bohemia.
The morphometric data were visualized using PCA in R 3.3.2 (R Core Team 2016), from package ade4 (Dray & Dufour 2007), the resulting graphics having been adjusted in CorelDraw 16. The variables for PCA were scaled and centred. The samples were labelled according to preliminary visual taxonomic assessment of herbarium material as a synthetic recognition of Valeriana officinalis var. tenuifolia-like plants, V. officinalis var. officinalis-like specimens and unclear cases. The hierarchical clustering (UPGMA, not displayed) was carried out in NTSYSpc 2.2, Exeter Software.

Morphometric analysis of the Valeriana officinalis dip loids
The material used for the morphometrics consists of a representative sample of a large metapopulation in the W part of the Elbe basin, C Bohemia (between Všetaty and Kostelec nad Labem), and a sample from another two C Bohemian populations (Hrabanov, Újezd u Průhonic); in all these subsamples, both forms are represented, although not at the same habitats. These samples were supplemented with herbarium specimens from the NE part of the distribution range of the form with narrow leaflets (Estonia and Saint Petersburg, Russia), and additional samples of (presumed) Valeriana officinalis var. officinalis s.str.
The PCA (Fig. 1) ordination shows a clear variation trend but with no really distinct entities, which corresponds to the hypothesis of two morphological units connected with a range of intermediates. The presumed identity of samples (if we disregard the intermediates) was obtained by means of a preliminary general habit assessment, and the two groups coincide with the main trend displayed in Fig. 1. Two major groups are also recognized by the hierarchical cluster analysis (UPGMA, not shown), with about 10 % of samples wrongly assigned in the preliminary assessment. The intermediates are almost evenly divided between the two major groups.

Brief characterization of the "pseudopratensis" form of Sebald
The morphometry-based subdivision of the diploids within Valeriana officinalis subsp. officinalis is in good agreement with the observations of Titz & Titz (1981) and Sebald (1996). If we summarize the data, the following short description of the "pseudopratensis" form can be compiled from the above sources and from our measurements: Diploids (2n = 14); early flowering (June; in the Scandinavian populations, peak flowering is in July); plants usually 50 -120 cm tall, without stolons; stem ± glabrous, with very short, thick hairs at or near nodes on ridges, usually suffused red in lower part (often not so in N part of range of this form); lower and middle leaves long petiolate, petiole usually villous in proximal part; leaflet pairs numerous, usually 6-9 in middle leaves; leaflets narrow, usually 2-6 mm wide, entire or rarely with a few teeth, ± glabrous, often with relatively sparse, very short and thick, almost subconic hairs on margins and some veins beneath (binocular lens needed).
Concerning distribution, the form is known to occur in Central Europe, from Switzerland, Austria, Slovakia and Hungary in the south, through Germany and Czech Republic to Denmark and S Sweden and along the S Baltic shore (Baltic states and the Saint Petersburg region). It is to be expected in Poland and Finland (specimens of promising appearance but with unknown chromosome number were seen). Willdenowia 47 -2017

Nomenclatural analysis
We have concluded that the diploid Valeriana officinalis consists of two population entities linked by intermediates; the rank of variety seems to be appropriate for this type of differentiation. As regards the tall, late flowering form with large, conspicuously dentate leaflets, it corresponds to the type of the name V. officinalis (Kirschner 2007;Kirschner & al. 2007; see also Fig. 2). It remains, then, to find the correct name for the form with narrow, entire leaflets, the "pseudopratensis" form of Sebald (1996). The first infraspecific names to be considered were published by Wibel (1799: 87). From the vicinity of Wertheim, Baden-Württemberg, Germany, Wibel published V. officinalis with two varieties (explicitly assigned to this rank): the first one, var. montana Wibel, characterized by narrower leaflets and the habitat confined to "montosis, saxosis et nemorosis", which makes it possible to equate this name with V. stolonifera subsp. an gustifolia Soó. We have to refrain from the typification of var. montana because the original material was not traced in the herbarium collections known to preserve Wibel's specimens (GOET, JE). The second variety, var. palustris Wibel, is characterized by larger, more conspicuously dentate leaflets, and may belong to V. excelsa Poir. or to V. officinalis var. officinalis.

The protologue of Valeriana officinalis var. tenuifolia
Enumeratio plantarum, vol. 2 (Vahl 1805) was published posthumously, under the editorship of N. Tönder, J. W. Horneman and P. Thonning. We compared the published protologue ( Fig. 3) with Vahl's original manuscript (deposited at Copenhagen, C) and found the two sources identical. Vahl (1805: 6) characterized Valeriana officinalis using his own phrase name (in principle, a Linnaean phrase name with an addition: "pinnis lanceolatis serratis"). Then [β] latifolia, without a new phrase name, quoting, together with several pre-Linnaean sources, the Linnaean phrase name from Species plantarum to the 13 th edition of Systema ve getabilium, and adding a reference to Oeder (1771: t. 570). Because of the inclusion of the phrase name used by Linnaeus in numerous works including Species plantarum (see Richter 1840), it is reasonable to equate var. latifolia Vahl with the typical variety of V. officinalis (Kirschner 2007), as a taxonomic synonym. It is also supported by the fact that plate 570 from Flora danica (Oeder 1771) was seen by Linnaeus and included in V. officinalis in his later works. The plate depicts a plant very similar to the epitype of V. officinalis (Kirschner 2007 ; Fig. 2). The second variety is V. officinalis var.
[γ] tenuifolia, published without any reference to other sources, solely on the basis of material seen by Vahl. The complete diagnosis of var. tenuifolia is quite concise. The entry in Vahl (1805) (Fig. 3) is closed by general remarks pertaining to the whole species and is therefore also applicable to var. tenuifolia. The following protologue characters are relevant from the point of view of the modern understanding of the variation within the V. officinalis group: plants 2 -4 feet tall; stem glabrous; stem leaves distinctly petiolate; petiole hairy (villous); leaflets linear, very narrow; leaflet margin entire; habitat: marshy groves ("in […] nemoribus paludosis").

Original material and typification of Valeriana officinalis var. tenuifolia
As stated above, in the protologue of this name, there is no reference to older sources nor any specimen or locality citation. The major part of the herbarium collection of M. Vahl is deposited in Copenhagen (C); the other herbarium with numerous specimens coming from the Vahl herbarium is G, but no relevant specimen was traced there. The keepers of the herbarium C searched the collection for Vahl's material of Valeriana officinalis s.lat. and found a single specimen that was studied and annotated by Vahl (Fig. 4). The specimen does not bear any text other than "Valeriana officinalis L. S. N. [= Linnaeus, Systema naturae] 44.5" on the reverse side of the sheet. However, the plant on the sheet matches the protologue characters of var. tenuifolia (see the list above) in every detail; in all likelihood the plant represents the original material for the name, which was also noted by I. Hiitonen on a separate label in 1966. As regards the absence of the epithet tenu ifolia on the specimen, it should be taken into consideration that Vahl's work was published posthumously. The above specimen is therefore selected as the lectotype of the name. Because of the utmost importance of the ploidy level in the V. officinalis group, and in order to avoid future misinterpretations of the name, we also select an epitype with determined chromosome number (diploid); see also the taxonomic discussion below.

Taxonomic interpretation of the name Valeriana officinalis var. tenuifolia
The lectotype and epitype (and numerous isoepitypes) belong to our "pseudopratensis" form characterized above. We therefore conclude that the correct name for the early flowering meadow diploid with narrow, entire leaflets within Valeriana officinalis is var. tenuifolia at the rank of variety and V. officinalis subsp. tenuifolia (Vahl) Schübl. & Mart. at the rank of subspecies. It is a mistake to use the latter name for the tetraploids of drier habitats, correctly to be called V. stolonifera subsp. angustifolia Soó at the rank of subspecies.

Notes on the epithet "angustifolia" in the Valeriana officinalis group
It was shown by Holub & Kirschner (1997) and Buttler & al. (2008) that the epithet "angustifolia" belongs to the earliest name at the rank of subspecies for what was commonly called Valeriana collina, but (due to the dou-

198
Kirschner & Zeisek: Diploids of the Valeriana officinalis group in Central Europe ble threat of homonymy) only outside V. officinalis (the basionym, V. angustifolia Tausch ex Host 1827, is a later homonym of V. angustifolia Mill. 1768, which applies to a taxon of Centranthus DC.). Within V. officinalis, the epithet cannot be used at the rank of subspecies because of the existence of two earlier, seemingly heterotypic names, usually referred to as V. officinalis var. angustifolia Hayne (1813) and V. officinalis var. angustifolia Wahlenb. The latter name was published (Wahlenberg 1824: 19) with a short diagnosis ("β. angustifolia, pinnis subintegerrimis"), with a direct reference to variety "γ" in Roemer & Schultes (1817: 351). On the page referred to, we find a variety "γ", but with the epithet tenuifolia and a reference to Vahl (1805) and another reference to Dufresne (1811, where var. tenuifolia was accepted without any reference). Thus, the name V. officinalis var. angustifolia Wahlenb., renaming the existing, legitimate V. officinalis var. tenuifolia, was nomenclaturally superfluous when published and thereby illegitimate (McNeill & al. 2012: Art. 52.1), is homotypic with var. tenuifolia (Art. 7.5) and at the same time is illegitimate as a later homonym (Art. 53.1) of V. officinalis var. angustifolia Hayne. The name published by Hayne (1813: t. 32 and text without pagination) refers to Schrader (1806: 85), where two varieties were recognized but their names were not in a form that could be validly published ("β. foliis angustioribus"). Hayne gave a number of references to various sources, including images, but there is a nice plate of var. angustifolia accompanying the description, with all the relevant details and characters, and that is suitable for the role of the lectotype (see below), when no herbarium specimens are extant. As regards the taxonomic interpretation of this name, the lectotype plate leaves no doubt that Hayne's var. angustifolia is to be equated with V. officinalis s.str.

A note on the name Valeriana altissima
Although the Code (McNeill & al. 2012) reckons with mistakes in newly published names (e.g. in Art. 41), there are mistakes in names that have nomenclatural consequences. One such mistake is the name Valeriana altissima Besser [recognized as a mistake by Koch (1840)]. Pohl (1809) published V. exaltata "Mikan jun.", a name introduced but not published by Mikan, and later treated and typified by Kirschner & al. (2007, equated  ) is treated as a separate species, V. altissima, by other authors. They therefore did not validly publish the name V. altissima because they did not accept it in their treatment (Art. 36.1(a)).
It was Besser (1821), in his enumeration of plants of Volyn, who validly published the name Valeriana altissima "Mikan". The ascribed authorship is here understood as an indirect reference, via "V. altissima" in Roemer & Schultes (1817), to V. exaltata J. C. Mikan ex Pohl (1809), who in fact published V. exaltata and V. sambucifolia, not V. altissima. Because of the indirect form of the references to both Roemer & Schultes (1817) and Pohl (1809), V. altissima Besser can only doubtfully be considered as a superfluous, illegitimate substitution of V. exaltata. Moreover, there is a V. altissima Hornem. (Hornemann 1815), a name published on the basis of plants cultivated in the Copenhagen Royal Gardens, with a diagnosis different from (though similar to) those of V. altissima Besser and V. exaltata, i.e. a name with independent typification. The name V. altissima was later used at the rank of variety, as V. officinalis var. altissima ("Mikan") W. D. J. Koch, synonymous but doubtfully homotypic with the earlier V. officinalis var. exaltata (J. C. Mikan ex Pohl) Kostel. The nomenclature of V. exaltata and V. altissima is summarized below.
Note. -There is a presumably authentic specimen of Valeriana exaltata (BM 1134437) sent by J. C. Mikan to J. J. Roemer and acquired by R. J. Shuttleworth. The transcribed label repeats the diagnosis from Pohl (1809), and it is difficult to decide if Mikan collected and sent the material in the period between 1804 (when the name was first introduced) and 1809, or between 1809 and 1819 (when Roemer died), which is more probable.

Selected synonyms of Valeriana officinalis var. officinalis
As a summary, we list the main synonyms of Valeriana officinalis var. officinalis. It should be noted that only the epitype of V. officinalis is identical with the neotype of V. exaltata, so that the two names are not homotypic (Kirschner & al. 2007).